As an enterprise employs more and remotely managed networks that include more and more devices and services, it becomes difficult to monitor performance in a manner that can keep pace with the speed at which and ways with which anomalies can escalate into problems. The enterprise might have tools with which to discover events after they happen. These tools, however, are insufficient for rapidly assessing how observed performance in one device or service may later impact performance of another device or service. As a consequence, a network manager of an enterprise network is often playing catch-up when faced with performance issues.